We have been able to eliminate
sweet spots, free up loudspeaker placement restrictions and create reproduced
sound fields of placed objects that stay put when you move and turn as in
reality. The resulting formats are not overly complex but new improved
loudspeakers are required. The new formats are completely compatible with the
surround formats and can in fact be used to correct some shortcomings of these,
particularly relating to speaker placement issues and sweet spot location –
that is if the listener wants sweet spot targeted reproduction, of course.
To achieve this, new rendering
equipment is also required. But we must remind ourselves that at this stage we
have only considered the direct sound field component from the sources. We now
need to consider the remaining sound field components.
What other sound
components are there?
The traditional framework for classification of sound types
considers the direct, early reflection and reverberant field components. These
are generally considered separate because of their differing behaviours. One
aspect not formally treated in the classification is to distinguish low
frequencies. We do this because of the lack of human
directional perception for low frequency sound means that all the low frequency
sound field components, - the direct, early reflections and reverberant sound
field parts can be considered as a group from the point of view of creating
placed sound objects.
This is not so with the medium to high
frequency components. Low frequency sound reproduction will be treated
separately in a later paper.
RECONSTRUCTING THE COMPLETE SOUND FIELD
Direct sound component
We understand the direct sound
field component. This is the first arrival sound and travels directly to the
listener from the sources. Given that VWF can place sound sources in space so
that the correct listener perspective is provided, the direct sound field
component is taken care of.
Early reflection component
The early reflection sound is
usually the second arrival, having been reflected by objects and boundaries in
the vicinity of the sources. What we know of the early reflections is that the
apparent source of each reflection is not the location of the original sources.
Each early reflection will be modified by specular (hard) and diffuse
(illuminatory) reflection components. The diffuse reflection component will
come from the point of impact on the reflecting surface. The specular
reflection will appear to come from an image point behind the surface. Both
reflection types will then interact with the directly radiating sound field
components to create colourations that characterise the capture environment and
each listener location. The characterization has both location and spectral
modification parts that relate to the properties of the reflective surfaces.
Given that with VWF we can
arbitrarily place sound sources at fixed locations in space, we can now place
each and every one of the reflection sources correctly at their real and
virtual locations. As we are also correctly placing each of the direct sound
sources in space, the resulting interacting sound field will behave correctly
for each and every listener location just as was the case in the original
environment for each location.
This does not necessarily mean that we have
solved the problem of early reflection reproduction. We still need to locate
the early reflection sources to behave as if we are listening in the capture
environment. Fortunately a number of factors assist here.
First, most capture auditoria have dominant
early reflections, and these are predominantly specular in nature. Accurately
placing the sources and so reproducing the first, say 10 early reflections may
well suffice.
Second, we could spatially characterise the
auditorium and so derive its spatial transfer function. This could be done once
and prior to any performance as it will largely be a characteristic of the
building and the general region that any sources (orchestra etc) will be
placed. This could be done by either actual measurement or by acoustic
modelling of the structure.
Third, if we have appropriate microphones,
these will automatically capture the location of all sources including the
early reflections [].
If a suitable microphone is used for the
recording, the set of all reflections “seen” by the microphone will be
correctly captured from the microphone vantage point and therefore will
automatically be correctly placed at reproduction. Unfortunately the result,
whilst very good will be less than ideal because the reflections will be those
“seen” at the microphone vantage point and will not correctly scale with
listener relocation in the listening environment at any other listener
location. All listeners bar the one at the position corresponding to the capture
location will receive “a view of a view”.
Another alternative is to model or measure
the early reflections themselves and then use post processing to correctly but
artificially place the early reflections at reproduction. By this means the
early reflections can be correctly placed or rendered to represent the original
set for all listeners regardless of location.
This approach would enable the
parameterising of both real and virtual recording environments and thus the
ability to recreate these environments at will.
Metadata (data about the venue acoustic
behaviour) could be used to distribute the characteristics of any desired venue
and the location of the microphones at the same time as the recorded sound
signals themselves.
This could be of great value to the gaming
and virtual reality markets. You could be exploring a virtual cave in darkness
and could sense approaching a wall, then, for a quick change, switch to the
intimacy of a phone booth, for example. Other treatments would also be possible.
Scene changes in movies and games could be accompanied by appropriate early
reflection acoustic behaviour changes.
Whilst accurate reproduction of the early
reflections in recording venues is now possible with VWF, in practice, we find
that for applications other than gaming and virtual environments, all but the
worst early reflections are completely overwritten by the reverberant field
components.
Reverberant field component
The medium to high frequency reverberant
component of the sound field is different. It is effectively “what is left
over” after the direct sound and the early reflections. The reverberant sound
field part is related to the sound from all the original sources but contains
no directional information. It gives the impression of the capture venue size
and nature through decay time constants and spectral acoustic colouring over
time. A different set of rules applies to correctly reproducing the reverberant
sound field than those for the direct and early arrival sound when listeners
are free to move and turn in the reproduction environment.
The reverberant sound field is the result
of many reflections from objects and surfaces in the original listening
environment. It must be related to the sum of all the sources present in the
recording environment, but is by definition characteristically lacking in any
discernible acoustic directional information for any listener location.
With a correctly reproduced reverberant
sound field, the listener should not sense any directional information when moving
around in the listening environment. This means that the source of the
reverberant sound field must be hard to find. An implication of this would be
that the placement of such a source would have to be non-critical.
The reverberant sound field will be
considered next.
The
information contained herein is copyright to HuonLabs. No material can be
reproduced in its totality or in part or without the express permission of
HuonLabs Pty Ltd. Any reference to this material must quote the HuonLabs
source. Trade marks and Patent applications apply to most aspects of the work
disclosed here. Contact HuonLabs for further details, product information or
licensing enquiries.
